Do agents really need blockchain payments?

Where is the Agent economy heading? Payment is only the starting point; on-demand, one-off markets are the new form of future finance. An in-depth interpretation of the three-stage evolution and disruptive imagination of the Agent Economy.

Author: Lao Bai

It just took me half an hour to finish reading this excellent article by @Leozayaat

It was a bit taxing on the brain cells, but it is very interesting thinking, especially the judgment about the future Agent Economy era

Previously, everyone always thought that in the economic activities of the Agent world, the best entry point for Crypto is payment. Over the past two years, there has even been a mainstream view in the circle - blockchain is not built for humans, but for Agents, because having a wallet is too easy, and small payments are too suitable

Just the past two days, @hosseeb from DragonFly posted a Tweet, including that he also said something similar at 2049. He believes his previous judgment was wrong: Agents do not need Crypto, it is we humans who need it

What he means by "do not need" here mainly refers to payment

Over the past year, Visa, Stripe, and other companies have all been actively positioning themselves in Agent payments. The vast majority of Agent payments are completed through traditional card networks and the financial world, while Crypto protocols such as X402 account for a negligible share

Agents do indeed need payment, but that does not mean they need an entirely new payment network

What if the new scenario of the Agent Economy is not payment, but something else entirely, in financial markets?

Leo's article is talking about this

Let me first briefly summarize the framework of the original article

"

The Agent Economy may go through three stages.

Stage 1: Agents buying from humans. Agents purchase services such as APIs, data, and computing power on behalf of humans. The core problem is how to enable Agents to pay autonomously.

Stage 2: Agents buying from agents. Agents begin to purchase services from each other, and the problem shifts from payment to fulfillment: how to ensure the other party delivers? The article proposes using Bond / Collateral to constrain Agents. For Agents that can create identities infinitely, reputation is easily abused, while collateral that can actually be confiscated is a more reliable constraint.

Stage 3: Disposable Markets. The most imaginative part.

For example: an Agent temporarily needs 1,000 hours of GPU computing power to complete a task, but worries that the supplier cannot deliver on time. It can create a market on demand, letting other participants price the specific risk of "whether the supplier can deliver on time," and use capital to bear the corresponding risk.

The market does not need to be listed in advance, nor does it need to recruit market makers in advance. As long as there is real demand, it can try to find capital willing to bear the risk.

This means that the financial market itself can become an on-demand infrastructure.

"

Let me talk about how I feel after reading it

For Stage 1 payment, X402 does currently have a bit of a "falsified" Feel, which is also the reason Haseeb proactively "admitted his mistake." Of course, the track has only just begun in its early stage, and it is too early to conclude that this thing will not work

For Stage 2 reputation, this is what ERC8004 is doing. Similar to X402, based on current data, it is also a track that leans toward being "falsified." Leo's original article has arguments for this view, so I will not elaborate in detail. Simply put - "having an on-chain identity does not equal having credible reputation; having a reputation record does not equal having real transaction history, and the cost for Agents to farm reputation is extremely low"

Stage 3 Disposable Markets - this is that Agents can create one-time markets on demand, price specific risks, and find capital willing to bear the risk.

This is a completely new economic form. Logically speaking, it Makes Sense in the silicon-based life world, because their search for, discovery of, decision-making about, communication about, and pricing of information are completely different from human society, and their efficiency is N orders of magnitude higher than that of humans

Whether it is really like this, no one knows. This is only a derivation, but at least it is a logically consistent derivation that opens up new imaginative space and possibilities

Especially with the recent launches of Grok Bot and Meta Muse, it feels even more likely to accelerate the evolution of the Agent Economy. By then, various new forms of financial markets will definitely emerge, and it will absolutely not be just payment

If you have time, I still recommend reading the original article. It discusses many interesting experiments in economics and finance conducted by foreign laboratories using Agents.